Low latency DDR3 ram to run at 1333mhz? Is there any and what benefits could be seen using it?

Had the wild idea of upgrading the ram but the question is to what?

So the question I ask now running at 1333mhz is there any ddr3 ram that I could run in the M15x at less than CAS 9 latency? For example could I put in either 1600 or higher memory in and then have it running at 1333 and a lower CAS lets say 7 or 8? I guess it would auto downclock and therefore use lower CAS too but perhaps I am wrong...

If this were possible would I see any performance improvements at all? I know next to nothing about this...

Cheers

 

You would just be wasting your money. Completely.

 

Lets ignore money at the moment. I know the idea is pretty uneconomical but would there be any performance increase?

I am more interested in the answer to the question than actually upgrading unless there is a benefit upgrading down the line when newer ram is cheaper.

 

If you were using an integrated gpu, then you might notice 5% increase in FPS during games (maybe more - they tend to vary on the scale, but overall its not noticeable too much).

But since you are not using an IGP, you wouldn't notice any difference whatsoever.
Paying a premium price by going from 1333 to 1600 is really not justifiable even if you had an IGP.
Lower latency of 1333Mhz would be recommended, but only when you are initially getting the laptop or are buying the RAM yourself at the beginning (in which case you can often get it for a cheaper price tag) or are swapping out the RAM entirely so you can get higher capacity one.
Post purchase switching is not really necessary.
You would essentially be throwing your money away on something that won't produce any tangible results.

 

this gives you all the number for the timings and ns for them. so you can figure it out on your own. Don't listen to people just saying no or giving you random answers. Check it out for yourself

DDR3 SDRAM - Wikipedia, the free encyclopedia

 

So a lower CAS can equate to a decent bandwidth boost!?

I did read somewhere that DDR3 1333 running at CAS 7 is equivalent to running at a much higher frequency something like 500mhz faster than CAS 9!

If this is the case considering the 920xm and HM55 chipset can't and won't run a higher frequency, running a lower CAS could really help open up system performance!

HopelesslyFaithful do you recommend any such upgrade? What would be viable? Cheers mate.

 

Yes that's right, when it downclocks it will use lower timings.

You can see the SPD table with CPUz, it will show you the timings used for each frequency.

You can even just buy thaiphoon burner and flash your ram to a higher frequency and lower timings if stable. Can change the voltage as well.

Not sure if you've read this thread yet:
http://forum.notebookreview.com/har...extreme-boost-1600-2133-beyond-jedec-xmp.html

My ram was 1333mhz and I bumped it up to 1600. I did try higher but it wasn't stable.
I also added XMP support. But my chipset doesn't support it (tried to hardcode into the bios to enable an XMP profile but it didn't work).

So I would suggest getting thaiphoon burner, it's a cheaper option than buying new ram and you will use many times.

 

This is great news. I just worked out that 1333mhz ram running at CAS 7 will be faster (10.5ns) than 1600mhz at CAS 9 (11.25ns).

This means bandwidth wise I can potentially run ram that is the equivalent speed of somewhere between 1600mhz and 1800mhz DDR3 ram at higher CAS levels.

The Thaiphoon burner program sounds interesting but am worried about frying something. If I get decent 1600mhz memory then and then force it back to 1600mhz would it work with the 920xm? I thought it was limited to 1333mhz?

 

not sure if i miss read what you siad but 1333 has the exact smae bandwidth no matter what ns you have....it changes how fast it can react. just like the internet 10mbps is 10mbps but 10mbps at 50ms is better then 10mbps at 100ms

btw thanks moral never knew that existed! When ever i build a desktop i am totally playing around with this!

 

It doesn't work that way. All you are measuring is the delay for the CAS latency. But that is not bandwidth and there are also zillions of other timings.

That doesn't tell you anything other than the JEDEC specifications for the ram. I'm not sure how you are supposed to figure out if there is a benefit to upgrading your RAM from that.

 

he wanted to know what is the best timings to get the best performance....that tells you.

1600 at 8-8-8 gets you 10ns 1333 at 7-7-7 gets you 10.5ns

also moral hazard told him how to use that program to redo your ram to a higher setting. so you could get 1600 at 7-7-7 if your ram is stable

 

How does that tell you what the (non-existent) real world performance benefit is?

 

What timings does 1866MHz need to beat out that 8-8-8 1600MHz you posted? Is there a formula that can be applied?


Nevermind I looked myself, CAS 9 with 1866 should be pretty fast then.

 

not going to argue with you....

 

Check this out for IGP performance in relation to CAS. Otherwise hard to really measure real-world improvement other than a slightly more responsive system, but I find faster RAM helps much better with that than tighter timings.

 

again all comes down to how your usage and requirements are. I honestly question if a 10ns comapred to a 15ns makes an actual difference...that is such a small amount of time i really do wonder. Also I don't know how any program needs 12GBps of transfer rate...I know some due but do normal people use that? For an example, Windows OS, gaming, and browsing? Also most laptops are single channel right? My old P4 had ~7GBps(dual channel) lol and my laptop gets only ~10GBps....seems stupid.

 

Most laptops are dual channel. Only some netbooks are single channel but now pretty much everything is dual channel. There really is no advantage except for RAM critical apps, of which there are few, that it will make any marked difference. System responsiveness is improved a little bit, otherwise there's very little value in it.

 

+1. I measured iGPU performance at http://forum.notebookreview.com/hp-...ebook-2560p-owners-lounge-21.html#post8585800 . When running in a single-channel configuration the higher clocked RAM did see perforamance approximate dual-channel levels. In the dual-channel configuration there was little benefit in going from 1333Mhz to 1866Mhz RAM.

 

any difference if u tried different latencies?

 

Upping the frequencies also decreases latency. Latency makes 0 practical difference.

 

Depends on what timing. Look at the chart

 

What "timing" does it depend on? The biggest increase was like 3% in a pure CPU synthetic benchmark, with the average difference being like 1%. And it made all of .5% difference for gaming.

 

...do i have to explain a chart that explains everything? you can have lower latency with a low bandwidth depending on your timing.... faster bandwidth doesn't always mean lower latency as you can see by reading the chart. Where do you get your number of 3% 1% and .5%...very interesting on where you would have found such numbers

 

This chart?

I guess you're going to have to explain it because every time the frequency increased, the latency decreased. Like I said, increasing the frequencies will decrease latency... The only way it's going to make latency higher is if you purposely cripple the timings which would just be dumb.

Check the 3dmark/PCmark/Crysis/Metro performance for the percents. You should be able to calculate the % differences yourself.

 

note the CAS latency. I know better CAS latency is better but I am curious is the correlation between CAS latency and cycle time. How do they affect each other. 1600 ram has better CAS latency than 1866 and 2133 but they have lower cycle latency so i am curious on how each pair up and whether it even matters and if so by how much


Also i didn't know the 1866 can go lower than this chart but according to your graph it can. I know nothing of the current specs of DDR3 ram besides Wikipedia and some bored and searching newegg. So I am basing it off of what I have read I see from yours 1866 can go lower which is cool.

 

CAS latency is usually measured in clock cycles. So even though 1600mhz CL9 and 1333mhz CL9 both have 9 cycles of latency, the cycles happen faster on the 1600mhz ram and as a result it has less latency.

Though you're right in the sense that if you go by the manufacturer's stock JEDEC settings, you will usually be forced to use looser timings at higher frequencies which will offset the the lower latency of the higher clock rate. However, that's because the the timings are increased, not because of the higher clock rate.

But that aside, as you can see from the Anandtech review there is very little difference in performance from just decreasing the latency.

 

btw latency is directly related to both. Both affect it. high freq and lower latency made better latency but fro stock settings 1600 is best but if you get that program moral posted you are golden i guess

 

Interesting ongoing discussion! So I guess the final answer to a ram upgrade would be - no point in getting faster ram or lowering the latencies. Just perhaps to upgrade to 16GB of ram if it were needed.

Good to know.

edit: what if I were to install this KHX1333C7S3K2/4G - Kingston HyperX 1333mhz DDR3 CL7 memory and then using thaiphoon tighten the timings to CL6 or even lower ? Worth it lol ?

 

You have CL9 RAM right now right? I'd estimate you'd see a 5% improvement in pure CPU programs and like 1-2% for gaming. Not worth it, imo.

 

5% is quite nice! So it would improve video encoding performance

Perhaps when it is cheaper I will get 1600 hyperx and then tighten the timings. Unfortunately the HyperX I listed above only comes in 2x2GB flavour.